Western blot, a method to separate proteins by size and analyse their relative expression levels, is a much maligned technique of molecular cell biology. The website PubPeer is flooded with evidence of manipulated Western blots, where gel lanes were inappropriately spliced, or where bands digitally duplicated or erased. Some even question the technology as such, since it is indeed mostly Western Blots (and other gels, like Northern Blots for RNA or Southern Blots for DNA or RNA PCR-amplification gels) which are flagged for image manipulation.

It is however not the technology to blame for all the rigging done with it, but the simple fact that it is image based. Anyone with a minimum of image analysis skills or a good eye for duplications can spot Western Blot manipulations. You do not need to be an expert in the technology or even a biologist, to find data rigging. This is exactly why a certain aberration in Western Blot integrity is often either overlooked or dismissed as incompetent nitpicking: the absence of proper loading controls. Just like it seems to be occasionally the case in the publications from the lab of Richard Moriggl, director of the Ludwig Boltzmann Institute Cancer Research and professor at the Medical University of Vienna, Austria. The Moriggl lab studies molecular signalling in cancer cells and tissues, hence its focus on the analysis of regulatory protein phosphorylation and how it changes under various conditions. Such analysis must be always supported by proper gel loading controls, which seems not always be the case here. A reader of my site contacted me with some examples of such inappropriate gel presentation in Moriggl papers, some of which he already posted on PubPeer.

Do the bands of total protein AKT match those of phosphorylated (pAKT) in this Moriggl paper Mueller et al, Diabetes, 2017? It seems they do not, which indicates the two separate gels are presented as one.  The paper is so fresh, original data must exist to answer the questions. Source: PubPeer

There, a number of Western Blots for a phosphorylated protein seem not to fit the corresponding loading controls provided in the figure. It is not a small issue, because the purpose of such gel analysis is to compare the levels of a certain phosphorylated protein among different samples, say control versus various treatment or a the effect of a treatment at different time points. This is why it is important to ascertain that exactly same amounts of protein-containing cell lysates are present in each gel lane. Here, one does not rely solely on prior protein concentration measurements, but also performs loading controls of the same gel, for a protein which levels are not expected to change in the course of treatments. In the case of phosphorylation analysis, one usually probes with specific antibodies for the same total, non-phosphorylated protein. Only when the equal loading is ascertained can one compare the changing levels of the phosporylation of your protein of interest, or in fact of the overall expression of any other protein you wish to analyse, using corresponding antibodies.

Key point here: it must always be the same gel. It is of no use to detect amazing differences in a protein expression or phosphorylation, while not having a proof that the gel was indeed equally loaded. To probe a different gel where same samples were re-run for an allegedly corresponding loading control does not really make much sense: no two gels are ever exactly same, and a lot of incidental and not so incidental mistakes can happen during the sample loading or the gel run or transfer. This is also why one can spot if the two images really belong to the same physical gel: the lane widths and band sizes and shapes are a give-away if they do not match. Indeed, the sacked cheater Irina Stancheva was caught not only on gel band duplications, but also on using inappropriate loading controls.

FIG B
In Kaltenecker et al Diabetologia, 2017  , blots for p-AKT and Akt do not seem to match, does the panel HSC70 belong to any of these 2 gels, or is it a 3rd one? the question is raised for both panels, “Liver” and “EWAT”. Comments here and in other figures are from the reader, who provided me with those.

Yet the loading controls are often seen as irrelevant. The fallen star of plant sciences Olivier Voinnet built his misconduct defence on his having used a loading control library, where he pulled out a gel image to fit the analysis he aimed to publish. A loading control duplication in a paper from Ronen Alon’s lab at Weizmann Institute was also explained with the use of such libraries (see here).  In any case, even if the loading control panel is found to be photoshopped or duplicated, even some established scientists will object: it is just a loading control! The main analytic Western Blot panels are perfectly fine! This is however absolute nonsense, from the scientific point of view.  Every Western Blot analysis needs a loading control, done for exactly same gel. Otherwise the entire protein expression analysis becomes utterly useless and one can skip using figures in scientific publications altogether.

fig-c.jpg
Also in Themanns et al Sci Reports, 2016, HSC70 served as loading control. But the bands appear too different for it to be the same gel.

Moriggl never replied to my emails. Hence I contacted the Medical University of Vienna with the evidence in his papers, in August 2017 (incidentally, one member of their Scientific Advisory Board is the radiologist and self-plagiarist Hedvig Hricak, from Memorial Sloan Kettering Cancer Center (MSKCC) in New York, read here). The Vice-Rector for Research, Michaela Fritz  replied to me the next day:

“Thank you for the information. After consultation with the University of Veterinary Medicine Vienna, the case will be taken up by the Ombudsman of the Vetmeduni, which has already contacted you”.

Indeed, the Ombudsperson of the Vetmeduni Vienna, Mathias Müller wrote to me in parallel:

“Your mail from 23 August 2017 to the management of the Medical University of Vienna was forwarded to the Ombudsoffice of the Vetmeduni. The issues are currently being discussed with the authors. The PubPeer presentations will be addressed with the publishers by submitting point-to-point comments”.

FIG D

In Javeheri et al Cell Death & Disease 2016, blots for signal HA-EWS/FLI1 and HSC-70 (upper figure) as well as for phosphorylated protein RB (pS780-RB) and total protein RB (lower figure) seem not to match. These gels seem to lack their correct loading controls.
The mismatched phospho- and total protein Western blots are not the only problem with this Javeheri et al 2016 paper from the Moriggl lab. Another loading control blot seems to be duplicated (Source: PubPeer), which should be easily checked given the recent publication of this paper and certain availability of original data.

Prior to writing this article, I asked the Ombudsman Müller for an update, but received no reply.  There are more inconsistencies in Moriggl papers, like this collaborative paper Gotthardt et al Cancer Discovery 2016 from the neighbouring Vienna lab of Veronica Sexl at the same Medical University Vienna. Interestingly, Sexl has other papers flagged on PubPeer and had already to correct one, namely Kollmann et al, Cancer Cell 2013.

FIG G
The loading control for total protein STAT5A/B does not seem to match its phosphorylated blot in   Gotthardt et al Cancer Discovery 2016

It is of course very likely that correct loading controls will be found and Moriggl will be able to correct his publications to show the matching pairs of gels. Those papers are very fresh after all, the data is certainly readily available. Otherwise readers would be forced to be sceptical about the scientific validity of these Western blot analyses. Without the correct loading control for the exactly same gel, to show the protein loading ratios, a celebrated difference in expression or phosphorylation of a regulatory protein might be nothing but an artefact.


Update 2.11.2017. Matthias Müller, ombudsman of the Vetmeduni Vienna, wrote me this email today:

“Dear Dr Schneider!
With regard to your e-mail dated 23.Aug.2017 to the Rectorate of the Medical University of Vienna, which has been forwarded to the Ombudsman of the Vetmeduni, I can inform you that the issues were clarified with the authors of the publications in Cell Death & Disease and Diabetes.
Replies were made to the PubPeer posts by means of ‘point-to-point’ statements and also sent to the editorial boards of Cell Death & Disease and Diabetes with a request to consider a publication of a corrigendum.
Both Editorial Offices have now approved the Erratum requests and will publish the Corrigenda in the earliest available journal issue”.

Update 13.11.2007. Is the Moriggl lab under investigation by Austrian Science Fund FWF?


 

 

Donate!

If you are interested to support my work, you can leave here a small tip of $5. Or several of small tips, just increase the amount as you like. Your generous patronage of my journalism, however small it appears to you, will greatly help me with my legal costs.

€5,00

 

6 thoughts on “On Western blot loading controls: lessons from Richard Moriggl lab

  1. In my opinion the WB is a disgraced old technique which alraedy has good alteranatives such as quantitative mass spectometry and ELISA (which results are not dependent of images). An alternative to cope with the WB technical problems would be the further development of automated WB technology

    Like

  2. In ELISA you have no information on the relative size of the protein, where sin WB you do. This can yield a lot of interesting information, as post translational modifications can alter the migration of the protein. Such changes may include phosphorylation (e.g., if you use a phospho site specific antibody, seeing multiple bands indicates that one should look for other phosphorylation sites or modifications, that may occur in tandem.
    So it is a great method, but like everything there is a great deal of craft and shortcuts should not be taken.
    I would agree that mass spectrometry should be the gold standard, though all to often when submitting a proteomics paper a reviewer asks for some of the discoveries to be ‘validated by WB’. Of course it should be the other way around.

    Like

  3. In my opinion:
    1. An accurate determination of protein content of the samples and good laboratory practices should allow the researchers to load the same amount of protein in each lane of WB.
    2. Work with sample duplicates in the same gel would allow researchers to test their own accuracy in the WB.
    3. To load different amounts of protein of the same sample in different adjacent lanes of the gel would allow researchers to test how ‘lineal’ is their effect.
    4.A loading control should not change during the experiment. This should be confirmed.
    5. When you measure ‘ratios’ of ‘activated’ vs a ‘non activated’ protein you are not measuring ‘activated’ vs ‘total the total protein you are studying. And the ‘total protein’ should be an amount much bigger than the ‘activated protein’ to have a sound measure.
    …..
    Just some basic rules of basic biochemical experiments that authors, reviewers and editors seem to have forgotten, and could avoid many fake results.

    Like

    1. I agree that mass spectrometry should be the gold standard and that after choosing the protein or proteins to study by proteomics we can further analyse these proteins by ELISA (including PTMs if you detect something by mass spectrometry) and also by WB (by following the basic experimental rules)
      Nevertheless I think the WB technique can be improved by automating it
      Having 3 different techniques at least to confirm the expression of a protein or proteins would be the ideal

      Like

  4. An Erratum in Diabetes for Mueller et al 2017,
    Diabetes 2017 Nov; db18er02a. https://doi.org/10.2337/db18-er02a

    “In the above-mentioned article, the following errors were detected:
    In Fig. 2C displaying AKT and phosphorylated AKT (pAKT) expression in
    epididymal white adipose tissue (eWAT) and liver, the AKT immunoblotting
    was not the corresponding blot to the pAKT. The authors are replacing this
    figure with the corresponding total AKT in the liver and providing a blot
    repeat of the eWAT, where all antibodies were probed on the same membrane.
    The authors recalculated this blot accordingly with similar findings to
    that previously published.
    In Fig. 6E, AKT and pAKT were not shown on the same gel. The authors
    repeated this Western blot and detected AKT and pAKT on the same immunoblot.
    The recalculated data are now shown in a revised Fig. 6E.
    In online Supplementary Fig. 4F, the authors now present the respective
    loading control, using Actin expression analysis together with that for AKT
    and pAKT.
    The authors apologize for any inconvenience caused to the readers. The online
    version has been updated to correct these errors”.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s